Michelle Bamberger and Robert E. Oswald (Photo credit: Marcellus Protest)
Extraction of hydrocarbon gas from tight shale formations using horizontal drilling and hydraulic fracturing has been advertised as a path toward energy independence for the United States and is being promoted worldwide. This is tempered by environmental and societal concerns that have led to banning the practice in some countries (e.g., France), at least one state in the U.S. (Vermont), and numerous towns and cities in the United States. In the United States, the process itself is largely regulated at the state level, with exemptions from federal laws regulating air, drinking water and hazardous waste disposal. Regulation at the state level varies considerably among states with significant shale deposits, as does the level of enforcement of regulations. The argument often given to suggest that the process is safe cites the fact that in the sixty years since the first gas well was hydraulically fractured, the industry has not found proof it finds acceptable that drinking water has been contaminated. This assertion is not universally accepted because of at least two factors.
First, it is based on a narrow definition of hydraulic fracturing, that is, solely the process of stimulation of the well; whereas, the public and many in academia are more concerned with the entire life-cycle of the drilling and extraction process with many possible routes of environmental contamination. The second issue is the burden of proof. Is it the public or the government that bears the burden of proving that environmental harm has occurred or should the industry be required to provide scientifically acceptable proof of the safety of the process? In this paper, we will discuss regulation briefly followed by a more detailed discussion of health effects of shale gas extraction, and possible impacts on food safety.
In a recent issue of Independent Science News (1), William Sanjour carefully reviewed the inherent problems of regulatory agencies. While his view was informed by thirty years of experience working for the US Environmental Protection Agency (EPA), his comments raise important concerns for state regulatory agencies as well. One important point that he raised was: “Agencies which enforce regulations should not write the regulations.” This was recognized by the US Department of the Interior after the Deepwater Horizon blowout when the Minerals Management Service, with allegedly strong connections to industry, was divided into permitting (Bureau of Ocean Energy Management) and regulatory (Bureau of Safety and Environmental Enforcement) branches.
Taking the example of New York State, the Department of Environmental Conservation (DEC), which has written the proposed hydraulic fracturing regulations, will be responsible both for issuing permits for the wells and enforcing the regulations. The two most important concerns are that the NYS DEC has been chronically understaffed, particularly with regard to permitting and regulating gas wells. Expansion of the program to high volume shale gas wells will only exacerbate the problem. Secondly, the agency has the dual missions of extracting minerals efficiently and protecting the environment. While an increase in state or federal bureaucracy will never be politically popular, there is an inherent conflict of interest between these two missions and it makes little sense to charge a small staff with this dual mandate. Careful oversight of this global industry, which is in the best interest of the public and the industry, can only be done if regulatory agencies are truly independent and have sufficient staffing to detect problems and provide effective enforcement.
We have noted previously (2) that health issues surrounding shale gas extraction have some similarities to that of tobacco. Whereas no medical or public health reasons exist for treating nicotine and its delivery system differently from any other drug that is regulated by the FDA, for political reasons, companies selling this product were not subjected to federal laws controlling drug sales. Likewise, no proof of the safety of the product was required. Instead, the burden of proof fell on government and academic scientists, who eventually demonstrated the dangers associated with tobacco use. Demonstrating the potential for the political process to work against the pubic interest, this carcinogenic product remains freely available and the adverse health consequences remain a burden on our medical system. In a similar manner, proponents of shale gas extraction have simply asserted that the process is safe, based on experience with low volume vertical wells. If held to this standard, one could imagine that Merck could have asked for approval of Vioxx® based on the safety record of ibuprofen or aspirin.
Our interest has been on the use of animals as sentinels of human health in areas experiencing extensive shale gas and oil extraction. Animals tend to have greater exposure to environmental threats than humans because they typically have greater exposure to air, soil and groundwater and have more frequent reproductive cycles and shorter generation times. For example, animal owners can often afford to purchase drinking water for their own use, but purchasing drinking water for a herd of cattle can be beyond the means of some farmers. Thus, the cattle have a much greater exposure to potentially contaminated water than the farmer.
The question is whether or not any of the processes associated with shale gas and oil extraction have led to adverse health consequences in animals or humans. We have published an analysis of case reports from 24 families with plausible exposures to toxicants associated with one or more phases of the drilling process (2). One approach to obtaining more definitive evidence might be to determine if specific chemical residues can be found in the environment following hydraulic fracturing operations that were not present before those operations began, identify a route of exposure, and then attempt to determine if the concentrations present in the exposure pathway could have caused health problems. On the surface, this is a logical approach that has been used for many years by environmental toxicologists.
But the problem faced in analyzing the health effects of shale gas extraction is not as simple as finding hexavalent chromium in the drinking water. It is an extremely complex problem with the potential of exposures from multiple chemical entities (known and unknown), multiple routes of exposure, and unknown interactions between chemicals. Furthermore, recent evidence suggests that the effects of chemicals at high doses cannot accurately predict effects at lower does (nonmonotonic dose response curves; 3). This is particularly true for endocrine disrupting chemicals that have been reported to be components of hydraulic fracturing fluid (4). For these reasons, the maximum contaminant levels (MCLs) of specific compounds that are used to determine the safety of drinking water are questionable. The problem is compounded by inadequate or nonexistent predrilling testing of air and water in many cases and by nondisclosure agreements tied to victim compensation that remove the details of specific cases from further study.
A good example of the approach described above is the extensive investigation of volatile organic compounds (VOCs) exposure by the Texas Department of State Health Services (TxDSHS) in Dish, Texas (5). Whereas this expensive and exceedingly well done study provided no evidence that residents were exposed to VOCs from shale gas operations, the authors note that this negative evidence represents only one time point. It does not rule out exposures at other times or locations. The authors wisely chose to include metabolites of VOCs, which extends the time scope of the study as the metabolites remain in the body longer than the parent VOCs. However, the fact remains that this study describes only ongoing or recent exposures at only one time point using a limited population and cannot really be used to prove that a process is safe.
Our view is that animals (including humans) can be used as biological integrators and that disease state can, if used properly, be a readout of toxicity. This might be used in studying effects of human and animal exposures as well as possible impacts on the food supply. In a first-pass investigation, this circumvents problems of multiple exposure pathways, multiple toxicants, nonmonotonic dose response curves, etc. The question is how to collect information and analyze it correctly. For this, we turn to the medical literature as one might view the problem as similar to, albeit more complicated than, a possible disease outbreak. One approach is to begin by analyzing individual cases to determine the characteristics of the disease and possible routes of exposure (2).
The idea is to use this information to inform further studies of prevalence and cause, as a large number of cases provides a stronger case for further studies. This is where the use of animals as sentinels of human health becomes important. A farm family living near a drilling operation provides only a limited number of cases. If the couple has reproductive problems, suggesting that the drilling operation is to blame is tenuous without many other cases showing similar problems. However, if this is a cattle farm with careful record keeping, it may be possible to show that, for example, one hundred heifers have reproductive problems that correspond temporally with drilling operations or specific violations (e.g., leak of a wastewater impoundment, well casing failure). Again, this does not provide definitive proof, but the increased sample size serves to guide further work regarding possible effects on reproduction by some aspect of the drilling operation.
One can take this a step further by analyzing more specific exposures. In several cases, we have found herds that have been split between two or more pastures (2). In those cases, the portion of the herd exposed to impoundment pond leaks, contaminated well water, or creeks into which wastewater was allegedly illegally dumped displayed reproductive failure while the unexposed portion of the herd was not affected. Likewise, this is not definitive evidence of cause-and-effect, but raises serious red flags and points the way to further studies.
Documenting individual cases will always be limited to individuals that are willing or able to talk to investigators but in our research we have run into significant unnecessary obstacles. For example in one small area of Pennsylvania, residents were surveyed as to whether drilling had impacted the perceived quality of their water supply (the water was supplied by wells between 40 and 200 feet deep). As of this writing, 50 of 132 families indicated that they noted changes in water quality (Prof. John Stolz, Duquesne University, personal communication) but only a small fraction were willing to provide further documentation.
In other cases, residents having presumed issues with water quality have signed nondisclosure agreements in exchange for compensation from industry. This is a worrying distortion of an important business practice. If we take the example used above of a pharmaceutical company, it would be considered unethical to compensate individuals suffering from an unknown side effect of a drug and then require them to sign nondisclosure agreements in an attempt to keep the drug on the market. The use of the nondisclosure agreement in issues relating to public and environmental health should be illegal (6).
In all of the cases that we have studied, the burden of proof for demonstrating harm has fallen squarely on the shoulders of those that have allegedly been impacted. Given that drilling has been concentrated in rural areas with low per capita income, the ability to prove harm is sometimes economically difficult or impossible. This brings up the subject of environmental justice. In New York State, for example, the New York City and Syracuse watersheds will be exempted from any drilling activity due to dangers inherent in the process. After securing this exemption, New York City’s mayor called for the expanded use of hydraulic fracturing to provide more energy for New York City. The division of New York State into zones of environmental privilege and zones of potential environmental sacrifice is contrary to any realistic notion of environmental justice–but is a likely consequence of current policies.
So What is the Next Step?
We believe that more careful epidemiological studies are needed. Despite our emphasis on animal health, public records are far more sophisticated for human than animal health. For example, milk production seems to have fallen in areas in Pennsylvania experiencing extensive drilling (7), but the analysis is at the level of individual counties, and the data are not available for an analysis at higher resolution. In a preliminary report, Hill (8) has found in human databases that babies born near drilling operations tend to have lower birth weights and reduced APGAR scores. Although details are not yet available, two large healthcare systems serving Pennsylvania and southern New York, Geisinger Health System and Guthrie Health, have embarked on a study of potential health effects of drilling using their patient database (9). The advantage of these epidemiological studies is that the databases were generated for other purposes and do not have biases associated with the drilling issue and the large datasets provide considerable statistical power assuming that proper controls are employed.
Analyzing health effects on the level of individual counties or similar coarse-grained geographical unit is relatively straightforward. However, more detailed studies that define risks within a certain distance from a source of pollution are more difficult. One of the problems in analyzing all of these data is defining the x-axis. That is, in principle, one might consider a gas drilling operation as a possible point source of pollution. However, the problem is not that simple and is becoming more difficult with time. If one considers Bradford County in Pennsylvania and maps all of the gas wells within a ten-mile radius of a given farm, it is not unusual to find five hundred or more wells (only a percentage of these would be high volume hydraulically fractured wells).
At this point, only about 2% of the total number of planned wells have been drilled in Bradford County, so defining the x-axis as the distance from a well (i.e., point source) is problematic and constantly changing. When we take this one step further, we realize that some of these wells have not had any problems, whereas others may have had a casing failure, a blowout preventer failure, or the leak of an impoundment pond. For this reason, any fine-grain analysis of health issues will have to be approached with care. Nevertheless, we believe that careful epidemiological studies are essential in order to define the prevalence of health problems that may be associated with drilling operations and perhaps to define the practices that are most problematic in terms of human and animal health. With careful epidemiological studies, it may be possible to identify particular risk factors in drilling operations. Of course, the downside of this approach is that those living in proximity to drilling operations (and those consuming food produced near them) become unwitting experimental subjects.
Whereas we believe that careful studies should proceed and continue to inform policy decisions, the question of how science can be used to make effective regulatory decisions is a topic common to many environmental and public health risks (10). The common gold standard of proof is statistically rejecting the null hypothesis with a probability of a false positive of 1 or 5%. One might start with a null hypothesis that states that proximity to a well pad has no correlation with an adverse health effect. But in the case of human hazards, minimizing false positives may be less important than minimizing false negatives. That is, if a correlation between an adverse health outcome and proximity to a well pad is found to be statistically not significant at the 0.05 level, perhaps having a p value of 0.06, then the null hypothesis that the adverse health outcome is unrelated to the proximity to a well pad would be accepted. This standard of proof is accepted practice in a laboratory situation because of the essential conservative nature of science, which is to avoid making incorrect conclusions at the expense of missing important effects. But laboratory science moves at a deliberate pace and the daily workings of an individual laboratory rarely affect the immediate quality of life for a large population.
In our drill pad example, however, the implications of accepting the null hypothesis may be much greater than the implications of rejecting it. Perhaps improving the survey instrument or taking a slightly larger sample could push the data to a significant result. Without these enhancements, we have a false negative, and an incorrect result. In this example, framing the null hypothesis is equivalent to defining the burden of proof, and the burden of proof does not fall upon the oil and gas industry. And we know that the lack of proof of harm is equated by politicians and policymakers with the lack of harm.
A more nuanced approach to analyzing environmental data is therefore essential to avoid a statistical bias against false negatives. Different levels of proof, which may be more appropriate for assessing any potential harm associated with drilling, are discussed in a report from the European Environment Agency (10). Using verbal descriptors, the levels of proof include “beyond all reasonable doubt,” “balance of evidence,” “reasonable grounds for concern,” and “scientific suspicion of risk.” By more carefully identifying the areas of ignorance and taking a realistic approach to false negatives, we believe that science can effectively inform public policy on this important issue.
The impacts of shale gas drilling on food safety have not been carefully studied and little is known. Again working from case studies, we have reported (2) that cattle exposed to wastewater from drilling operations have been taken to slaughter without further testing. In many instances, those that have died following exposures to drilling operations have been taken to rendering plants, where their flesh has been processed into feed for other animals (e.g., chickens, pigs, fish). These are concerns without definitive proof. No routine testing is done before allowing the products derived from animals exposed to environmental contaminants to enter our food supply, nor is there adequate information on appropriate hold times (time between exposure and slaughter) for production animals with known exposures to environmental toxicants. Equally, nothing is known about impacts on vegetable and fruit crops. We have visited farms with crops growing within tens of feet from produced water tanks (i.e., tanks containing fluids that return from the well along with the gas) and well-heads, and fields of corn and squash near impoundment ponds that contain wastewater at different stages of the drilling process. The extent to which this is a concern is not known, but deserves careful study.
The unconventional gas-drilling boom has swept across the globe in recent years without evidence that environmental and public health can be protected. In the United States, the industry enjoys extensive subsidies, which include, among many others, exemptions from federal laws regulating clean air, clean water, and the disposal of toxic substances. A patchwork of state regulations allow secrecy rather than disclosure of substances used in all steps of the process, and nondisclosure agreements have been used to block access to information on specific cases that could provide meaningful public health information. Without complete transparency (disclosure of all chemicals used and outlawing nondisclosure agreements in cases involving public health) and complete testing, science cannot proceed unimpeded. Without careful science demonstrating, not the absence of proof of harm, but rather the clear absence of harm to public health, neither state nor federal regulations can assure that the food supply and the health of individuals living near gas drilling and processing operations will be protected.
Until we can protect public health with greater certainty, unconventional shale gas extraction should be severely limited or banned, using the subsidies currently provided to support this industry to instead develop and deploy renewable forms of energy.
1. W. Sanjour, “Designed to Fail: Why Regulatory Agencies Don’t Work.,” 2012, (accessed October 12, 2012).
2. M. Bamberger and R.E. Oswald, “Impacts of Gas Drilling on Human and Animal Health,” New Solutions, A Journal of Environmental and Occupational Health Policy 22(1) (2012): 55-77.
3. L. N. Vandenberg, et al., “Hormones and Endocrine-Disrupting Chemicals: Low-Dose Effects and Nonmonotonic Dose Responses,” Endocr Rev 33(3) (2012): 378-455.
4. T. Colborn, et al., “Natural Gas Operations from a Public Health Perspective.,” International Journal of Human and Ecological Risk Assessment 17 (2011): 1039-1056.
5. Texas Department of State Health Services, “Dish, Texas Exposure Investigation, Dish, Denton County, Texas,” 2010, (accessed October 12, 2012).
6. K. Franklin, “Confidentiality Agreements. The Problem: Confidentiality Agreements in Lawsuit Settlements Can Be Harmful, Even Deadly, to the Public,” 2007, (accessed October 12, 2012).
7. R. Adams and T.W. Kelsey, “Pennsylvania Dairy Farms and Marcellus Shale, 2007-2010,” 2012, (accessed October 12, 2012).
8. E.L. Hill, “Unconventional Natural Gas Development and Infant Health: Evidence from Pennsylvania,” 2012, (accessed October 2, 2012).
9. ”Guthrie Health and Geisinger Collaborate on Marcellus Shale Research Effort,” 2012, (accessed October 12, 2012).
10. P. Harremoës, et al., “Late Lessons from Early Warnings: The Precautionary Principle 1896-2000,” 2002, (accessed October 28, 2012).